The Last Mile: Where AI Stops and Expertise Begins in Streaming, with Michael Chan and Edit Kovacs

How far can AI actually take you in a running streaming service? In this episode, Accedo's Edit Kovacs, VP of Europe and Latin America, and Michael Chan, VP of Operations North America, make the case for forward deployed engineering in the AI era. While AI has made the first 80% faster than ever, the last mile, the challenges from device fragmentation, DRM compliance, recommendation engines that need continuous care, and still demands something no model can provide: engineers with deep domain knowledge and enough scar tissue to know what breaks in production and why.The build-and-handover model is no longer fit for purpose. The future belongs to long-term embedded partnerships focused on continuous operation, evolution, and outcomes.Timeline: 01:15 What is forward deployed engineering and why now?04:15 Why AI has shifted the bottleneck from models to implementation05:32 The last mile problem in complex streaming environments08:50 Device fragmentation and the ceiling AI can't lower11:30 How AI changes the relationship between client and tech partner17:18 Maintaining context in long-term embedded engagements21:34 Domain expertise, regional compliance and the 20% that matters28:40 The future of FDE: from building to governing agentic workflows